<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
Whisper Small GL - Santiago Paramés-Estévez
This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3179
- Wer: 15.2334
Model description
This model was fine-tuned using Sanchit Gandhi's notebook: https://huggingface.co/blog/fine-tune-whisper
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0707 | 2.69 | 1000 | 0.2596 | 16.4915 |
| 0.0063 | 5.38 | 2000 | 0.2952 | 15.8583 |
| 0.0014 | 8.06 | 3000 | 0.3105 | 15.2624 |
| 0.0011 | 10.75 | 4000 | 0.3179 | 15.2334 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
